src/hg/utils/automation/HgAutomate.pm 1.27

Index: src/hg/utils/automation/HgAutomate.pm
===================================================================
RCS file: /projects/compbio/cvsroot/kent/src/hg/utils/automation/HgAutomate.pm,v
retrieving revision 1.26
retrieving revision 1.27
diff -b -B -U 4 -r1.26 -r1.27
--- src/hg/utils/automation/HgAutomate.pm	8 May 2009 23:05:21 -0000	1.26
+++ src/hg/utils/automation/HgAutomate.pm	8 Jun 2009 18:40:49 -0000	1.27
@@ -36,9 +36,9 @@
       ),
     # Hardcoded paths/commands/constants:
     qw( $gensub2 $para $paraRun $centralDbSql $cvs
 	$clusterData $trackBuild $goldenPath $images $gbdb
-	$splitThreshold $setMachtype
+	$splitThreshold $runSSH $setMachtype
       ),
 );
 
 #########################################################################
@@ -197,9 +197,9 @@
   # If it doesn't produce a recognizable uptime result, return a
   # very high load.
   my ($mach) = @_;
   confess "Must have exactly 1 argument" if (scalar(@_) != 1);
-  my $cmd = "ssh -x $mach uptime 2>&1 | grep load";
+  my $cmd = "$HgAutomate::runSSH $mach uptime 2>&1 | grep load";
   verbose(4, "about to run '$cmd'\n");
   my $load = `$cmd`;
   if ($load =~ s/.*load average: (\d+\.\d+).*/$1/) {
     return $load;
@@ -214,10 +214,10 @@
   # swarm and hgwdev are now valid workhorses since they have access to hive.
   confess "Too many arguments" if (scalar(@_) != 0);
   my %horses = ();
   foreach my $machLine ('swarm', 'kolossus', 'hgwdev',
-	`ssh -x encodek parasol list machines | grep idle`,
-	`ssh -x memk parasol list machines | grep idle`) {
+	`$HgAutomate::runSSH encodek parasol list machines | grep idle`,
+	`$HgAutomate::runSSH memk parasol list machines | grep idle`) {
     my $mach = $machLine;
     $mach =~ s/[\. ].*//;
     chomp $mach;
     $horses{$mach} = &getLoadFactor($mach) if (! exists $horses{$mach});
@@ -314,9 +314,9 @@
   foreach my $paraHub (keys %cluster) {
     my $clusterInfo = $cluster{$paraHub};
     next if (! $clusterInfo->{'enabled'});
     next if ($onlySmallFast && $clusterInfo->{'gigaHz'} < 2.0);
-    my @machInfo = `ssh -x $paraHub parasol list machines | grep -v dead`;
+    my @machInfo = `$HgAutomate::runSSH $paraHub parasol list machines | grep -v dead`;
     my $idleCount = 0;
     my $busyCount = 0;
     foreach my $info (@machInfo) {
       if ($info =~ /idle$/) {
@@ -325,9 +325,9 @@
 	$busyCount++;
       }
     }
     my $batchCount =
-      `ssh -x $paraHub parasol list batches | grep -v ^# | wc -l`;
+      `$HgAutomate::runSSH $paraHub parasol list batches | grep -v ^# | wc -l`;
     my $expectedPortion = 1 / (1 + $batchCount);
     my $oomph = (($idleCount + ($busyCount * $expectedPortion)) *
 		 $clusterInfo->{'gigaHz'});
     &verbose(3, "chooseClusterByBandwidth: " .
@@ -480,12 +480,12 @@
 #########################################################################
 # Hardcoded paths/command sequences:
 use vars qw( 	$gensub2 $para $paraRun $centralDbSql $cvs
 		$clusterData $trackBuild $goldenPath $images $gbdb
-		$splitThreshold $setMachtype
+		$splitThreshold $runSSH $setMachtype
 	   );
 use vars qw( $gensub2 $para $paraRun $clusterData $trackBuild
-	     $goldenPath $gbdb $centralDbSql $splitThreshold );
+	     $goldenPath $gbdb $centralDbSql $splitThreshold $runSSH );
 $gensub2 = '/parasol/bin/gensub2';
 $para = '/parasol/bin/para';
 $paraRun = ("$para make jobList\n" .
 	    "$para check\n" .
@@ -506,8 +506,12 @@
 # as opposed to "scaffold-based" (no split tables; multi-level directory for
 # per-seq files, or use set of multi-seq files).
 $splitThreshold = 100;
 
+# ssh command and its options, the extra -o options prevent asking
+# questions about adding machines to known hosts
+$runSSH = "ssh -x -o 'StrictHostKeyChecking = no' -o 'BatchMode = yes'";
+
 $setMachtype = "setenv MACHTYPE `uname -m | sed -e 's/i686/i386/;'`";
 
 #########################################################################
 # General utility subroutines:
@@ -569,9 +573,9 @@
   my ($dbHost, $db) = @_;
   confess "Must have exactly 2 arguments" if (scalar(@_) != 2);
   my $query = "select genome,description,sourceName from dbDb " .
               "where name = \"$db\";";
-  my $line = `echo '$query' | ssh -x $dbHost $centralDbSql`;
+  my $line = `echo '$query' | $HgAutomate::runSSH $dbHost $centralDbSql`;
   chomp $line;
   my ($genome, $date, $source) = split("\t", $line);
   return ($genome, $date, $source);
 }
@@ -581,9 +585,9 @@
   my ($dbHost, $db) = @_;
   confess "Must have exactly 2 arguments" if (scalar(@_) != 2);
   my $query = "select scientificName from dbDb " .
               "where name = \"$db\";";
-  my $line = `echo '$query' | ssh -x $dbHost $centralDbSql`;
+  my $line = `echo '$query' | $HgAutomate::runSSH $dbHost $centralDbSql`;
   chomp $line;
   my ($scientificName) = split("\t", $line);
   return ($scientificName);
 } # getSpecies
@@ -593,9 +597,9 @@
   # does not.
   my ($mach, $file) = @_;
   confess "Must have exactly 2 arguments" if (scalar(@_) != 2);
   confess "undef input" if (! defined $mach || ! defined $file);
-  my $count = `ssh -x $mach ls -1 $file 2>>/dev/null | wc -l`;
+  my $count = `$HgAutomate::runSSH $mach ls -1 $file 2>>/dev/null | wc -l`;
   chomp $count;
   return $count + 0;
 }